Vol. 523: 81–92, 2015 MARINE ECOLOGY PROGRESS SERIES Published March 16 doi: 10.3354/meps11196 Mar Ecol Prog Ser

FREEREE ACCESSCCESS Calcareous spherules produced by intracellular symbiotic bacteria protect the Hemimycale columella from predation better than secondary metabolites

Leire Garate, Andrea Blanquer, María-J. Uriz*

Centre d’Estudis Avançats de Blanes (CEAB-CSIC), Access Cala Sant Francesc 14, 17300 Blanes (Girona), Spain

ABSTRACT: Benthic sessile organisms in general, and in particular, have developed an array of defense mechanisms to survive in crowded, resource and/or space-limited environments. Indeed, various defense mechanisms may converge in sponges to accomplish a defensive function in an additive or synergetic way, or to operate at different times during the sponge’s life cycle. Moreover, sponges harbor highly diverse microbial communities that contribute in several ways to the host’s success. Although some symbiotic bacteria produce chemical compounds that protect the sponge from predation, the possible deterrent function exerted by the calcareous coat of a sponge’s endosymbiotic bacterium has not, to date, been explored. Hemimycale columella is an Atlanto-Mediterranean sponge, which produces bioactive metabolites and has been reported to host an intracellular bacterium with a calcite envelope. Calcibacteria accumulate in high densities at the sponge periphery, forming a kind of sub-ectosomal cortex. They have been suggested to provide the sponge with several benefits, one of which is protection from predators. In this study, we assess the relative contribution of the endosymbiotic calcibacteria and bioactive compounds produced by H. columella to defend the sponge against sympatric predators. Deterrence experi- ments have revealed that the sponge combines >1 defense mechanism to dissuade a large array of potential predators; this represents an example of the evolutionary fixation of redundant mech- anisms of defense. The chemicals deterred Paracentrotus lividus, Chromis chromis, Oblada mela- nura, and Diplodus vulgaris, but not Parablennius incognitus and Coris julis, while the spherules of the symbiotic calcibacteria significantly deterred all predators assayed.

KEY WORDS: Chemical defenses · Calcifying bacteria · Sponge endosymbiosis · Sponge deterrence · Calcite spherules · Hemimycale columella · Atlanto-Mediterranean

Resale or republication not permitted without written consent of the publisher

INTRODUCTION Benthic sessile organisms in general, and sponges in particular, have developed an array of defense Species coexistence, which determines the biodi- mechanisms to survive in crowded, resource and/or versity of a given ecosystem, is the result of several space-limited environments. Structural materials, such long discussed, biological, and ecological mechanisms as external and internal skeletons, dermal spines, or such as environmental variation (Chesson & Warner protruding spicules serve as defense for benthic in - 1981), resource and/or niche partitioning (Chesson vertebrates and fish by protecting their soft tissues 2000), and species-specific interactions, which involve and, thus, dissuading most potential benthic preda- species-specific mechanisms of defense (Buss 1976). tors. Conversely, bioactive chemicals usually act in a

*Corresponding author: [email protected] © Inter-Research 2015 · www.int-res.com 82 Mar Ecol Prog Ser 523: 81–92, 2015

less generalist way. Some chemicals may deter one or Freeman 2012). For instance, some sponge symbiotic more species from predation on the producer organ- bacteria produce chemical compounds that protect ism while they may not deter others (Becerro et al. the sponge from predation (Thacker et al. 1998, 2003). On the whole, chemical defenses have been Haber et al. 2011, Esteves et al. 2013). However, reported to significantly contribute to the structure the possible deterrent function of an endosymbiotic of sponge assemblages on reefs (Loh & Pawlik bacterium, other than that mediated by bioactive 2014) and have been proposed to favor complex inter- chemicals, has not been explored to date. action networks, which are responsible for in creasing Hemimycale columella (Bowerbank, 1874) is a com- species coexistence and thus biodiversity (Buss 1976, mon encrusting demosponge (Order Poecilo sclerida) Loh & Pawlik 2014). On the other hand, mineral skele- widespread in the Mediterranean and North Atlantic tons would only improve species persistence by offering sublittorals. The species, which has a reduced ecto - general protection to the organisms (Uriz et al. 2003). somal skeleton (Van Soest 2002), produces chemical Sponges are a notable component of the marine compounds with cytotoxic and antimitotic activities benthos, where they share habitat with an array of (Uriz et al. 1992, Becerro et al. 1997a) that might deter potential predators (McClintock et al. 1994, Wulff its potential predators. However, H. columella has also 2000, Santos et al. 2002, Leon & Bjorndal 2002, been reported to host an intracellular bacterium with Knowlton & Highsmith 2005). Thus, besides compet- calcifying abilities (Uriz et al. 2012). Bacteria that are ing for growth space with other benthic organisms surrounded by a 100 nm thick calcite envelope have such as algae, , ascidians, and bryozoans, been detected by catalyzed reporter deposition fluo- sponges must also handle predation. Indeed, sponge rescence in situ hybridization (CARD-FISH) and trans- survival has required the development of several mission electron microscopy (TEM) in high numbers defense mechanisms, which comprise the production within a particular sponge cell type called calcibac - of chemical compounds and structural elements (Uriz teriocyte (Uriz et al. 2012). Thousands of bacterium- et al. 2003, Jones et al. 2005), along with cryptic produced calcite spherules are accumulated at the growth habits (Bertolino et al. 2013). sponge periphery, forming a kind of sub-ectosomal Many of the bioactive compounds produced by cortex that mimics a rudimentary exoskeleton (Uriz sponges with antimitotic, cytotoxic, antibacterial, et al. 2012). It has been proposed that this unusual, and/or antitumor activities (Amade et al. 1987, Uriz intimate symbiosis, which is vertically transmitted to et al. 1992, Becerro et al. 1994, Monks et al. 2002, progeny and constantly present in all populations of H. Sipkema et al. 2005, Blunt et al. 2009, Hardoim & columella examined along the western Me diterranean Costa 2014) inhibit the settlement of foreign larvae in sublittoral zone, purportedly provides the sponge with the proximity of the sponge (Martin & Uriz 1993, several benefits, among which protection from pre - Becerro et al. 1997a, 2003, De Caralt et al. 2013) or dators has been highlighted (Uriz et al. 2012). deter predation (Uriz et al. 1996, Ribeiro et al. 2010, In this study, we aimed to assess the relative contri- Arias et al. 2011). Spicules have also been reported to butions of the calcite spherules of endosymbiotic cal- dissuade sponge predators to some extent (Burns & cibacteria and the bioactive compounds produced by Ilan 2003, Hill et al. 2005, Jones et al. 2005, but see H. columella to sponge defense, and whether the Chanas & Pawlik 1995). Thus, various defense mech- combination of secondary metabolites and the cal- anisms converge in most sponges to accomplish a cibacterial calcareous envelope exerted a synergetic defensive function in an additive or synergetic way, effect in deterring potential predators from feeding or to operate at different times during the sponge’s on the sponge. With this aim, we conducted several life cycle (Uriz et al. 1996). However, the opposite is deterrence experiments, both in the laboratory and also true: multiple functions have also been reported in the sponge’s habitat, with an array of sympatric for a sole defense mechanism (Thacker et al. 1998, potential predators ( and fishes). Becerro et al. 1997a). Furthermore, the efficiency of a deterrent mechanism can vary according to the predator (Becerro et al. 2003), which makes the MATERIALS AND METHODS results of deterrence assays difficult to generalize. To add to the complexity of defense mechanisms, Sampling, sponge identification, and location of sponges harbor highly diverse microbial communi- bacteria ties (e.g. Blanquer et al. 2013), which form stable symbiotic associations and contribute in several ways Between 8 and 12 individuals of Hemimycale col- to the host’s success (Taylor et al. 2007, Thacker & umella were randomly collected from the Blanes lit- Garate et al.: Chemical defenses versus endosymbiotic calcibacteria in sponge deterrence 83

toral zone, NW Mediterranean (41°40.12’N, 2°47.10’E), were initiated. Once the experiments were com- in each of 3 sampling dives, to prepare the artificial pleted all individuals were taken back to their natu- food used in the experiments. The species was taxo- ral habitat. nomically identified by phenotypic characters, such as external morphology (i.e. thick encrusting shape, presence of characteristic rounded pore sieves with Chemical extraction elevated rims, pale orange to pink color, and spicule shape [strongyles to styles], size [320 to 461 μm × 2.5 Ca. 25 g of fresh sponges, corresponding to 40 ml in to 7.5 μm), and plumose arrangement [Van Soest volume (according to the water volume displaced 2002]). The sponge samples were taken to the labo- when submerged in a measuring cylinder), were ratory in hermetic, seawater-filled bowls, blended, freeze-dried for 72 h and pounded. Acetone was used and weighed after removing foreign material under for chemical extraction because it has been reported a stereo-microscope. Half of the sponge mix was to extract a wide range of secondary metabolites frozen for obtaining the crude chemical extract (Cimino et al. 1993). The extraction was done in an (potential chemical defenses), while the other half ultrasound bath for cell breaking and was performed was used to isolate the calcite spherules, which in 2 steps. First, we added 20 ml of acetone per represented the purported physical defenses. gram of sponge powder, and the extraction lasted Light-microscope pictures were obtained by using for 25 min. Once the supernatant was removed, we forceps to break up recently collected sponges and added another 20 ml of acetone per gram sponge and by direct observation of the resulting disaggregated extracted it for 10 min. The supernatants from the 2 cells though a Zeiss (Axioplan) microscope con- extractions were pooled together in a previously nected to a Jenoptik/Jena (ProgRes C10 plus) digital weighed tube, and the solvent was totally evaporated camera. in a hood. The tubes were weighed again after dry- For TEM, samples of ca. 2 mm3 in size were fixed in ing to determine the amount of crude extract ob-

1% OsO4 and 2% glutaraldehyde (1:3) in 0.45 M tained. The procedure was repeated 4 times (25 g of sodium acetate buffer (pH 6.4) with 10% sucrose fresh sponge each time) and ended with a total crude (Leys & Reiswig 1998) for 12 h at 4°C. After rinsing in extract of 72 mg (0.45 mg ml−1 sponge), which was the same buffer, dehydration, and inclusion in preserved frozen in the darkness until the artificial Spurr’s resin, ultrathin sections were prepared, food was prepared. stained with uranyl acetate and lead citrate, and observed with a TEM (JEOL 1010), implemented with a Bioscan system (Gatan) for image digitaliza- Isolation of calcibacteria spherules tion (Microscopy Unit of the Scientific and Technical Services of the University of Barcelona). The presence and abundance of calcibacteria in For scanning electron microscopy (SEM), samples the sampled sponges (i.e. bacteria surrounded by a were fixed in a cocktail (6:1) of a saturated solution of calcareous coat) were confirmed through optic and

HgCl2 and 2% aqueous solution of OsO4 (Johnston electron microscopes. The calcibacteria coats, which & Hildeman 1982), cryofractured in liquid N2, de- are calcium carbonate made according to X-diffrac- hydrated, gold palladium metalized, and observed tion analysis (Uriz et al. 2012), were obtained directly through a Hitachi S-3520N SEM (Microscopy Service from fresh sponge samples. Ca. 25 g fresh sponge, ICM-CSIC, Barcelona). 40 ml in volume, were disaggregated and homoge- For experiments in the laboratory, the target pred- nized in sterile seawater to avoid dissolution of the ators were the Paracentrotus lividus and calcite-made, calcibacteria spherules. The whole pro - the fish Parablennius incognitus, which share habitat cess of spherule isolation consisted of a series of with H. columella. These 2 predators were collected centrifugations and re-suspensions (Fig. 1E,F) in an in sufficient numbers from the sponge habitat attempt to be as exhaustive as possible. Siliceous (Blanes littoral, NW Mediterranean; 41°40.12’N, spicules precipitated first, forming part of the pellet 2°47.10’E), transported to the laboratory in seawater after centrifugation, and were discarded. containers, and placed in an open-system aquarium The spicule-free homogenates were initially cen- at a similar temperature to that in their habitat trifuged at 200 rpm for 1 min (Step 1), and the super- (22°C). All individuals were from the same size-class natants with the calcite spherules were removed and (adults), and no male livery was shown by any of kept apart. The pellets, which still contained spherules, them. They were starved for 7 d before experiments according to light microscope observation, were re- 84 Mar Ecol Prog Ser 523: 81–92, 2015

Fig. 1. Calcibacteria in (A−D) Hemmycale columella, and (E,F) the process of calcibacterial isolation. (A) Sponge cells (cal- cibacteriocytes) containing the calcified bacteria (which appear refringent through the light microscope). (B) Huge amounts of calcibacteria after cell dissociation of fresh sponges (light microscope). (C) Scanning electron microscope image of a cryofrac- tured sponge showing entire and broken calcibacteria—i: internal side showing the nanospherules that form the calcite coat; o: outside of the calcite coat (arrows point to the zones where nanospherule arrangement in a layer is more conspicuous). (D) Transmission electron microscope image of a calcibacteriocyte containing 2 calcibacteria within their respective vacuoles (ar- rows); the calcareous envelope was dissolved during the fixation process by acidic fixators, i.e. glutaraldehyde). (E) Sponge homogenate after spicule removal: Step 1 of the cell dissociation and centrifugation process (see ‘Materials and methods: Isolation of calcibacteria spherules’). (F) Step 2 of the process in which the debris of most sponge cells has already been removed—a: supernatant containing isolated calcibacteria in suspension; b: layer of calcibacteriocytes; c: settled calcibacteria Garate et al.: Chemical defenses versus endosymbiotic calcibacteria in sponge deterrence 85

suspended in 7 ml of sterile seawater and centrifuged either the crude extract or the calcibacterial blend again at 500 rpm for 2 min (Step 2). The resultant was poured into 8 Petri dishes (6 plates for experi- supernatants were removed and set apart. The upper mental trials and 2 as hydration controls). The treat- layer of the pellet, which contained entire calcibacte- ment combining the crude extract and the calcibacte- riocytes, was also removed with a pipette, and re- ria was prepared by adding both 5.3 mg of sponge suspended with RIPA buffer (Sigma) and sterile crude extract and 48 ml of concentrated calcibac - seawater (1:1) to lyse the calcibacteriocytes; this was terial suspension to the carragenate−alga mix. After centrifuged at 500 rpm for 2 min (Step 3). The various cooling, the plates were removed from the Petri supernatants containing spherules were pooled to - dishes and weighed immediately before being offered gether and centrifuged at 2000 rpm for 4 min in order to the sea urchins. Two plates per treatment were to precipitate the calcibacteria spherules (Step 4). kept in aquaria free of sea urchins to estimate the pos- The spherule-free supernatant (verified through sible weight gains because of carragenate hydration. light micro scopy) was discarded, and finally the pel- let was re-suspended in 1 ml of sterile seawater. The whole process was repeated 3 times totaling 48 ml of Parablennius incognitus calcite spherules, which represented a concentration of 0.4 ml spherules ml−1 sponge. Hundreds of ca. 3 mm long, 1 mm thick pellets—an appropriate size considering the mouth size of the target fish—were hand made from smashed bread. Artificial food preparation Either the sponge crude extract solution or the spherule suspension was added in appropriated vol- Two types of artificial food were prepared accord- umes to bread pellets to obtain ecologically relevant ing to the feeding behavior of the target predators: con centrations (i.e. similar to those present in the 4% carragenate plates for sea urchins and bread sponge tissues). The treatments considered for P. pellets for fishes. incognitus were: sponge crude extract (chemical treat- ment), calcibacteria spherules, and acetone control. The chemical treatment was prepared by adding Paracentrotus lividus ca. 4 mg of crude extract, dissolved in 12 ml acetone, to 10 g (ca. 40 ml of bread pellets, measured in a The food controls were prepared by adding 120 g measuring cylinder) to obtain a concentration of ca. of the fresh alga Cystoseira mediterranea, which is 0.45 mg of crude extract per milliliter of bread pellets. part of the diet of P. lividus’ (Verlaque & Nédelec The spherule treatment was prepared as described 1983, Verlaque 1984), to 120 ml of 4% carragenate. above by adding 16 ml of spherules, suspended in To detect any deterrent effect of the solvent used in 1 ml of seawater, to 10 g, ca. 40 ml, of bread pellets the chemical treatment, acetone controls were also (resulting in a concentration of ca. 0.4 ml of spherules prepared by adding 12 ml of acetone to 120 ml of a per milliliter of pellets). The acetone control was pre- 4% carragenate−alga mixture (i.e. 2.25 ml of acetone pared by adding 12 ml of acetone to 10 g (40 ml) of per carragenate plate). Either the sponge crude ex- bread pellets. The pellets containing the treatments tract (chemical treatment) or the calcibacterial treat - were then air dried to facilitate manipulation. In the ments were added to 120 ml of slightly warm 4% previous experiment on sea urchins no differences carragenate seawater−alga mix. were found between the carragenate and acetone For the chemical treatment, we re-dissolved ca. controls, so we only considered the acetone control in 5.3 mg of crude extract in 12 ml of acetone and added subsequent experiments. this solution to the 120 ml of 4% carragenate−alga mix, which approximately mimicked the crude volu- metric concentration in the sponge (ca. 0.45 ml of In situ sympatric fish assemblage crude extract per milliliter of carragenate). For the calcibacterial treatment, we added 48 ml of The artificial food for the in situ experiment with the concentrated spherule suspension to 120 ml of a the sympatric fishes at the sponge habitat was similar 4% carragenate−alga mix, which approached the to that prepared for the fish experiment in the labora- calcibacterial density estimated in fresh sponges tory (see above) but the pellet size was larger (4 mm (0.4 ml of spherules per milliliter of carragenate). A long and 1 to 2 mm thick) in order to adapt the food total of 15 ml of the carragenate−alga mix containing to the mouth size of the fishes targeted. We per- 86 Mar Ecol Prog Ser 523: 81–92, 2015

formed the same 3 treatments (crude extract, cal- Coris julis. Thus, we recorded the behavior of these cibacteria spherules, and acetone control) as in the P. 4 fishes with respect to the food offered. The number incognitus experiment. Treatments were offered to of fish participating in the experiment, as estimated the fish assemblage at random. Fishes at sea were from the number of pellets that they ate or rejected, adapted to feed on artificial food offered by divers was >20 per species (see Table 3), although we were for 7 d prior to the experiment. unable to ensure that a given individual participated only once in the experiment. The artificial food was taken to sea in large plastic P. lividus experiment syringes (1 treatment−1) as described by Becerro et al. (2003). Treatments and controls (5 pellets treat- The sea urchins were starved for 1 wk and then ment−1) were randomly offered to fishes by slowly placed in individual 5 l aquaria, with continuous aer- releasing the pellets into the water. Two independent ation at 22°C. Three treatments consisting of (1) the SCUBA divers recorded the number of eaten or sponge crude extract, (2) the calcibacteria spherules, rejected pellets. A pellet was considered rejected by and (3) both components combined were offered. We a fish if tried and spat out 3 or more times. When a used a total of 30 individuals, 6 per treatment (includ- pellet was ignored, or tried by a fish just once or ing controls). Two other aquaria were disposed under twice and spat out and ignored, the outcome was the above conditions to sink 2 plates of each treat- annulled and a new pellet of the same treatment was ment to assess their increase in weight due to hydra- offered later. tion during the experiment. The plates were ran- domly distributed between individuals, and the experiment lasted for 48 h. The plates were then Comparative deterrence quantification recovered, slightly toweled, and weighed to calcu- late weight losses that were due to sea urchin graz- A deterrence index (Becerro et al. 2003) was used ing, after discounting the mean increase in weight of for comparing sea urchin and fish deterrence in the 3 plates used to control hydration. experiments. The index (DET) was defined as: EC – ET DET = OC OT (1) EC P. incognitus experiment OC where EC is either the number of control pellets eaten After 1 wk of adaptation to aquarium conditions, by fishes or the weight losses in the control agar each P. incognitus individual was placed in a 5 l plates offered to sea urchins and OC is the number of aquarium (N = 14) with continuous water flow at a control pellets offered or the initial weight of the con- constant temperature (22°C). The experiment lasted trol plate; ET is the number of treatment pellets eaten for 8 d. Every 2 d we offered 10 pellets of each treat- or the decrease in weight of a treatment plate and OT ment (crude extract, spherules, and acetone control) is the number of treatment pellets offered or the ini- in random order to 4 randomly selected fishes, and tial weight of a give treatment plate. DET varies from recorded the number of pellets eaten or rejected per 0 (no deterrence) to 1 (total deterrence). treatment. No pellet was ignored when offered to fish in this experiment. At the end, we had a total of 16 replicates per treatment. Those fishes that were not Statistical analyses involved in a given trial were fed daily ad libitum with Sera® marine granulate. Data from the experiments in the laboratory on the sea urchin P. lividus and the fish P. incognitus were analyzed by 1-way ANOVA after rank transforma- Sympatric fish experiment tion, since they did not meet the assumptions for parametric analyses. The significance values of the The field experiment was carried out in the Blanes post hoc pairwise comparisons (Newman-Keuls test) sublittoral zone (NW Mediterranean; 41°40.12’N, were adjusted by the false discovery rate (FDR) cor- 2°47.10’E), in summer 2013, on a rocky, 10 to 15 m rection for multiple comparisons (Yekutieli & Ben- deep, bottom. The most frequent fish species co- jamini 1999). occurring in the sponge habitat were Chromis Data from the sea experiment were analyzed using chromis, Diplodus vul garis , Oblada melanura, and log-linear models for contingency tables. We tabu- Garate et al.: Chemical defenses versus endosymbiotic calcibacteria in sponge deterrence 87

lated our data with treatment (control and treated 4.5 food), fish species tested, and consumption (eaten or 4.0 rejected) as factors, and the number of occurrences 3.5 (pellets) in each category as observed cell frequen- 3.0 cies (Sokal & Rohlf 1995). The statistical significance of the deviations of the observed frequencies from 2.5 the expected frequencies was evaluated by Pearson’s 2.0 chi-squared. Statistical analyses were performed with 1.5

Statistica 6 software. Ingested food (g) 1.0

0.5

RESULTS 0.0 C AC CE CS CECS Treatment Calcibacterial presence/abundance Fig. 2. Deterrent effect of treatments assayed on the sea urchin Paracentrotus lividus (N = 6; C: carragenate control; Calcibacteria were present in the Hemimycale col- AC: acetone control; CE: sponge crude extract; CS: calcibac- umella sponges used for the experiments, as proved teria spherules; CECS: crude extract+calcibacteria). Hori- by light and electron microscope observations. Cal- zontal bars at the top of the panel indicate no significant cibacteria spherules were extraordinarily abundant differences between treatments after false discovery rate correction (p < 0.021) (Fig. 1A−D) in the sponge homogenates either free in suspension (due to their small size [<1 μm in dia - meter] and low weight) as a result of calcibacterio- trols (carragenate control and acetone control), cyte damage or on the upper layer of the pellets which were eaten similarly (p = 0.22). There were no within denser entire calcibacteriocytes. SEM pictures significant differences in feeding between the calci - of cryofractured sponge tissue showed calcibacteria bacteria spherules and the chemical treat ment (p = with a 100 nm coat of nanospherules arranged in a 0.29), or between the chemical treatment and the layer and with inner material corresponding to the chemical+calcibacteria spherule treatment (p = 0.29; bacteria (Fig. 1C). Images of intracellular bacteria Fig. 2). Thus, the crude extract, the calcibacteria deprived of the calcareous coat (likely due to calcium spherules, and the calcibacteria+crude extract simi- carbonate dissolution during the pH-lowering fixa- larly deterred sea urchins from feeding on H. col- tion process) were obtained by TEM. The intracel - umella, but the latter combination did not deter the lular vacuoles maintained the size and shape of the sea urchin in an additive or synergetic way. calcibacterial coat (Fig. 1D).

P. incognitus experiment P. lividus experiment ANOVA on the number of pellets ingested by P. ANOVA results on ranks showed a significant incognitus showed significant differences among effect (p < 0.001) of treatments on ingested food treatments and the control (Newman-Keuls test, p < (Table 1). Post hoc comparisons (Newman-Keuls test) 0.001; Table 2). Post hoc multiple comparisons after after FDR correction proved significant (p < 0.001) FDR correction showed significant differences (p < differences between the 3 treatments and the 2 con- 0.001) between the calcibacterial treatment and the control but not (p = 0.13) between the chemical Table 1. One-way ANOVA on the deterrent effect of several treatment and the control (Fig. 3). Thus, only the treatments (carragenate control, acetone control, crude ex- tract, calcibacteria spherules, and chemical and bacterial Table 2. One-way ANOVA on the deterrent effect of several components combined) on the sea urchin Paracentrotus treatments (control, crude extract, and calcibacteria spherules) lividus on the fish Parablenius incognitus

Effect SS df MS F p Effect SS df MS F p

Treatment 1653.00 4 413.25 17.37 <0.001 Treatment 3548.49 2 1774.25 25.49 <0.001 Error 594.50 25 23.78 Error 2853.01 41 69.59 88 Mar Ecol Prog Ser 523: 81–92, 2015

12 nificantly less often than the control for 2 out of the 4 assayed fishes (Table 3; χ2, p < 0.001). 10

8 Comparative deterrence quantification 6 The deterrence index (DET), which represents the 4 relation between the consumed food and the food offered in treatments and controls, varied between

Number of pellets eaten 2 treatments and among species (Fig. 4). It approached 1 for both the chemical and the calcibacterial treat- 0 C CE CS ments in Paracentrotus lividus, while it significantly Treatment varied between the calcibacterial (DET = 0.64) and Fig. 3. Deterrent effects of treatments assayed on the fish chemical (DET = 0.1) treatments for P. incognitus. Parablennius incognitus (N = 16; C: control; CE: sponge The fish deterred most by the 2 treatments (DET = crude extract; CS: calcibacteria spherules). Horizontal bars indicate no significant differences between treatments after 1, crude extract and calcibacteria spherules) was false discovery rate correction (p < 0.023). Boxes represent the sparid Oblada melanura. Conversely, the labrid ±SE, bars ±SD Coris julis, the pomacentrid Chromis chromis, and the sparid Diplodus vulgaris showed significantly spherules of the symbiotic calcibacteria defended H. lower deterrence indices for the chemical treatment columella against predation by the small sympatric than for the calcibacterial treatment (DET = 0, DET = fish P. incognitus. 0.4, and DET = 0.4, respectively) (Fig. 4).

Sympatric fish experiment DISCUSSION

The 3-way log-linear model for the contingency The various deterrence experiments performed table with treatment, fish species, and ingested revealed that the sponge Hemimycale columella food as factors indicated that the assayed fish spe- combines >1 defense mechanism to dissuade poten- cies, which shared habitat with the target sponge, tial predators. Some predators are deterred by both were differently deterred from feeding by the 2 secondary metabolites and calcibacteria—a case treatments assayed (Table 3; χ2, p < 0.001). There example of the evolutionary fixation of redundant were high significant differences (p < 0.001) mechanisms of defense in a species to widen the between the calcibacterial treatment and the con- spectrum of predators deterred. On the other hand, trol for the 4 sympatric fishes. Conversely, the chemical (crude extract) treatment was eaten sig- Calcibacteria 1 spherules Crude extract Table 3. Frequency table from the in situ sympatric fish assemblage experiment that was used for con tingency table analysis (143 pellets 0.8 were offered per treatment). Asterisks indicate deterrent effect on ingestion (χ2, *p < 0.05; **p < 0.01; ns: non-significant) 0.6

Ingested pellets Chromis Oblada Diplodus Coris Total 0.4 chromis melanura vulgaris julis

Deterrence index 0.2 Acetone control Yes 31 33 27 40 131 0 No 4 6 1 0 11 Crude extract Yes 23 0 17 43 83 Coris julis No 22ns 24** 14** 0ns 60 Calcibacteria spherules Chromis chromisOblada melanuraDiplodus vulgaris Yes 0 4 35 39 Paracentrotus lividus No 38** 42** 18** 7* 105 Parablennius incognitus Total 118 105 81 125 429 Fig. 4. Deterrence index of sponge crude extract and calci- bacteria spherules for several species assayed Garate et al.: Chemical defenses versus endosymbiotic calcibacteria in sponge deterrence 89

sponges are not the only organisms to present 2 dif- et al. 1987, Uriz et al. 1992, Becerro et al. 1997a). Here ferent types of defenses; crude extracts and sclerites we report on another defensive function of these have also been reported to exert anti-predatory func- secondary metabolites since they discourage the sea tions in gorgonians (van Alstyne & Paul 1992). urchin P. lividus from grazing. It has also been re - The spherules produced by the symbiotic calcibac- ported that P. lividus is deterred by the crude extract teria significantly deterred all species assayed; thus of the sponge Crambe crambe (Uriz et al. 1996, they appear to represent a generalist defense mech- Becerro et al. 1997b) and the seagrass Posidonia anism. Conversely, the chemical extract of H. col- oceanica (Vergés et al. 2007), while it appears to con- umella deterred some of the species assayed, but sume the alga Caulerpa taxifolia during the months not others; thus it seems to represent a more species- when it presents the lowest amount of secondary specific defense mechanism. meta bolites (Lemee et al. 1996). Protection from the The sea urchin Paracentrotus lividus was deterred devastating grazing by sea urchins (Guidetti & Dulˇci´c from feeding on H. columella by both the sponge’s 2007) seems to be widespread among many benthic chemical extracts and the calcibacteria spherules, as organisms, which have developed deterrent toxicants. well as by a combination of both components; this The outcomes of the fish experiments differed finding agrees well with the observed lack of pre - according to the fish species assayed. In general, cal- dation of P. lividus on H. columella in the field careous spherules deterred fishes more efficiently (L. Garate, A. Blanquer, M.-J. Uriz pers. obs.). Several than sponge crude extract did, but, for some species, studies reported that some sponge components de- both components were similarly deterrent. Previous terred this sea urchin, but the outcome of the assays studies reported that sponge skeletal structures are performed here varied as a function of the sponge deterrents for fishes (Burns & Ilan 2003, Jones et al. species, the types of defense analyzed, and the sea 2005, but see Chanas & Pawlik 1995, 1996). H. col- urchin species used. Sponge spicules, spongin, colla- umella is a spicule-poor species, and the high con- gen, or calcium carbonate may deter some sea centration of calcibacteria spherules at the sponge urchins from predation (Pennings & Svedberg 1993, periphery may replace spicules as deterrent elements Uriz et al. 1996). Conversely, other sea urchin species for fishes. feed on sponges habitually, despite the presence of The indexes formulated to compare the deterrence siliceous spicules (De Ridder & Lawrence 1982, San- intensity among the potential predators assayed (DET) tos et al. 2002). These contrasting results illustrate varied across species depending on the treatment. predator-dependent outcomes to the same type of The calcibacteria DET index showed the highest defense. value for C. chromis and O. melanura, followed by P. lividus has been reported to feed on sponge the sea urchin P. lividus (DET = 1), and exerted the species devoid of spicules when food resources are lowest effect (DET = 0.19) on C. julis. Such differ- limited (Boudouresque & Verlaque 2007). Since H. ences may be due to differences in the habitual prey columella shows a relatively poor spicule comple- preferentially targeted in the field by each predator. ment, predation by P. lividus on this sponge species Thus, fishes such as C. julis, which usually feed on would be expected, but has not been observed. Our invertebrates provided with an external skeleton, results showed that the calcium carbonate spherules such as mollusks, gastropods, bivalves, and crusta - of the symbiotic calcibacteria at natural concentra- ceans (Kabasakal 2001, Stergiou & Karpouzi 2002), tions deter this sea urchin. The spherules may be may be more adapted to encountering calcareous unpalatable to sea urchins, but not strictly toxic structures in their diet. (Birenheide et al. 1993), and, likely, their high con- The DET index for the chemical treatment varied centration in sponge tissues may decrease sponge drastically with the predator species: while it reached nutritional quality; thus, field sea urchins may select its highest value for the sea urchin P. lividus and the other more attractive food sources for optimal fish Oblada melanura, and a medium value for D. growth. Moreover, it has been reported that both vulgaris and C. chromis, it was close to zero for P. calcite and aragonite deter some herbivore fishes incognitus and C. julis. from feeding (Pennings & Svedberg 1993), which has The contrasting deterrent effects found for the been related to a decrease in fish gut pH impairing crude extract may also be related to differences in food digestion (Schupp & Paul 1994). the natural feeding habits of the species assayed. O. Besides the symbiotic calcibacteria, H. columella melanura and D. vulgaris are considered opportunis- produces secondary metabolites with demonstrated tic predators that feed on an array of both benthic antimitotic, cytotoxic, and antibacterial activity (Amade and pelagic organisms (Pallaoro et al. 2003, 2006). 90 Mar Ecol Prog Ser 523: 81–92, 2015

Thus, they may select other, non-toxic food sources studies in which the possible synergism between in the field. The small bleniid fish (P. incognitus) cap- structural defenses and crude extracts from sponges tures small benthic in the field, while graz- has been considered showed disparate results (e.g. ing the surface of rocky substrata and encrusting Hay et al. 1994, Burns & Ilan 2003, Hill et al. 2005, invertebrates (Goldschmid & Kotrschal 1981); thus, it Jones et al. 2005, Ribeiro et al. 2012). might be adapted to ingest small amounts of po - The Hemimycale−calcibacteria symbiosis is not the tentially toxic species such as H. columella and C. only case in which a bacterium protects a sponge from crambe (Becerro et al. 1997b) while capturing small predation. Recently, a symbiotic beta-proteobacteria invertebrates dwelling on sponges. The labrid C. of the sponge C. crambe has been reported to partici- julis is a voracious species that has been reported pate in the metabolic pathways (Croué et al. 2013) of 2 to predate on crustaceans and gastropod mollusks highly deterrent metabolites of C. crambe (Uriz et al. (Fasola et al. 1997, Kabasakal 2001); apparently it 1996). On the other hand, polyketide synthethases also tolerates, to some extent, the bioactive com- (PKS) of bacterial origin, with bioactive functions, pounds produced by the alga Caulerpa prolifera have been found in many sponge−bacteria symbioses (Sureda et al. 2006). Thus, the 2 latter fishes seem to (Piel et al. 2004, Haber et al. 2011, Esteves et al. 2013). show some resistance to the secondary metabolites of In all cases, the resulting substances produced by benthic invertebrates. On the other hand, pomacen- symbiotic microorganisms, either secondary metabo- trid fishes such as C. chromis are also opportunistic, lites or carbonate spherules, may be used by sponge omnivorous species that include sponges in their diet species to their own benefit (deterrent, antibacterial, (Emery 1973, Emery & Thresher 1980, Horn 1989). C. or antifouling roles; Uriz et al. 1996), thus promoting chromis, however, has been reported to avoid artifi- the persistence of sponge−bacteria associations. cial food containing the crude extract of the nudi- Since the concentrations (at a volumetric propor- branch Discodoris indecora, which obtains its meta - tion) of chemicals and calcibacteria used in the ex - bolites from Ircinia spp. sponges (Marin et al. 1997). periments were roughly similar to those found in nat- Reinforcing the invertebrate periphery by mineral ural sponges, we are confident that protection from materials in order to make it less attractive to po - predation is one of the benefits that H. columella tential predators is the main function of mineral exo - receives from its symbiosis with calcibacteria. Its con- skeletons (Uriz 2006). Sponges concentrate micro - tribution to sponge survival may have helped estab- scleres at the periphery to form a mineral cortex lish this unique symbiosis between marine sponges (Uriz et al. 2003). Rohde & Schupp (2011) reported a and calcifying bacteria. However, the sponge’s higher deterrent effect of artificial food containing secondary metabolites also exert a deterrent effect siliceous spicules from the sponge cortex than from against some potential sponge predators. The calci - the choanosome. This is likely related to the spicule bacteria and crude extracts together do not seem to sizes and may depend on the mouth size of the preda- have an additive or synergetic effect on potential tor considered. Small spicules (microscleres) are sponge predators, but rather may be engaged in densely packed in the sponge cortex (Boury-Esnault enlarging the array of potential predators deterred. & Rützler 1997), and thus more likely to deter small The chemical defenses of H. columella contribute to predators, while protruding choanosomal mega scleres the complexity of Mediterranean species interactions, (from hundreds of micrometers to millimeters) likely which supports the theory of Buss (1976), which was deter larger mouthed predators (Uriz et al. 2003). The recently substantiated by Loh & Pawlik (2014) for calcified calcibacteria of H. columella are spherules sponge communities of coral reefs. In contrast, sym- of ca. 1 μm size that are transported by amoeboid biotic calcibacteria unambiguously contribute to pro- cells (calcibacteriocytes) to the sponge sub-ectosomal tect the sponge from generalist predators and thus zone (Uriz et al. 2012) where they form a kind of cal- favor the species’ success. This is the first time that a careous cortex. Since these calcareous spherules ap- physical defense produced by symbiotic bacteria has pear to be so efficient in deterring the potential pred- been documented. ators assayed, their high concentration at the sponge periphery may make them very efficient in deterring Acknowledgements. We thank Jan Sureda for help with the an array of small-mouthed predators. field experiments and A. Lloveras and G. Agell for help with Although chemical extracts and calcibacteria deterred the laboratory assays. This study (M.J.U.) has partially been funded by the projects MarSymbiomics (MINECO, I+D+I some potential predators individually, the deterrent of Excellence, CTM2013-43287-P), the Benthic Ecology effect did not increase in additive or synergistic ways Consolidate Team of the Generalitat de Catalunya (2014 when they were combined in a treatment. The few SGR-120), and BluePharm Train FP7 People-INT (Ref. 2013- Garate et al.: Chemical defenses versus endosymbiotic calcibacteria in sponge deterrence 91

667786). All experiments complied with institutional, na - Cimino G, Crispino A, Madaio A, Trivellone E, Uriz MJ tional, and international ethics guidelines concerning the (1993) Raspacionina B, a further triterpenoid from the use of animals in research. None of the species used were Mediterranean sponge Raspacionina aculeata. J Nat listed as ‘Endangered’. Prod 56: 534−538 Croué J, West NJ, Escande ML, Intertaglia L, Lebaron P, Suzuki MT (2013) A single betaproteobacterium domi- LITERATURE CITED nates the microbial community of the crambescidine- containing sponge Crambe crambe. Sci Rep 3: 2583 Amade P, Charroin C, Baby C, Vacelet J (1987) Antimicro- De Caralt S, Bry D, Bontemps N, Turon X, Uriz MJ, Banaigs bial activities of marine sponges from the Mediterranean B (2013) Sources of secondary metabolite variation in Sea. Mar Biol 94: 271−275 Dysidea avara (Porifera: Demospongiae): the importance Arias J, Santos-Acevedo M, Newmark-Umbreit F (2011) of having good neighbors. Mar Drugs 11:489−503 Evaluación del potencial antidepredador de los extractos De Ridder C, Lawrence JM (1982) Food and feeding mecha- orgánicos crudos de quince esponjas marinas. Bol Investig nisms: Echinoidea. In: Jangoux M, Lawrence JM (eds) Mar Costeras 40: 293−308 nutrition. AA Balkema, Rotterdam, p 55−116 Becerro MA, Lopez NI, Turon X, Uriz MJ (1994) Antimicro- Emery AR (1973) Comparative ecology and functional bial activity and surface bacterial film in marine sponges. osteology of fourteen species of damselfish (Pisces: J Exp Mar Biol Ecol 179: 195−205 Poma centridae) at Alligator Reef, Florida Keys. Bull Mar Becerro MA, Turon X, Uriz MJ (1997a) Multiple functions Sci 23:649−770 for secondary metabolites in encrusting marine inverte- Emery AR, Thresher RE (1980) Biology of the damselfishes. brates. J Chem Ecol 23: 1527−1547 Bull Mar Sci 30: 145−328 Becerro MA, Uriz MJ, Turon X (1997b) Chemically-mediated Esteves AI, Hardoim CC, Xavier JR, Gonçalves J, Costa R interactions in benthic organisms: the chemical ecology (2013) Molecular richness and biotechnological potential of Crambe crambe (Porifera, Poecilosclerida). Hydrobio- of bacteria cultured from Irciniidae sponges in the north- logia 355: 77−89 east Atlantic. FEMS Microbiol Ecol 85: 519−536 Becerro MA, Thacker RW, Turon X, Uriz MJ, Paul VJ (2003) Fasola M, Canova L, Foschi F, Novelli O, Bressan M (1997) Biogeography of sponge chemical ecology: comparisons Resource use by a Mediterranean rocky slope fish of tropical and temperate defenses. Oecologia 135: 91−101 assemblage. Mar Ecol (Berl) 18:51−66 Bertolino M, Cerrano C, Bavestrello G, Carella M, Pansini Goldschmid A, Kotrschal K (1981) Feeding ecology of three M, Calcinai B (2013) Diversity of Porifera in the Medi - populations of Blennius incognitus Bath, 1968 (Pisces: terranean coralligenous accretions, with description of a Teleostei: Blenniidae) during the reproductive period new species. ZooKeys 336:1−37 and under human influence. Mar Ecol (Berl) 2:1−14 Birenheide R, Amemiya S, Motokawa T (1993) Penetration Guidetti P, Dul i J (2007) Relationships among predatory and storage of sponge spicules in tissues and coelom of fish, sea urchins and barrens in Mediterranean rocky spongivorous echinoids. Mar Biol 115: 677−683 reefs across a latitudinal gradient. Mar Environ Res 63: Blanquer A, Uriz MJ, Galand PE (2013) Removing environ- 168−184 mental sources of variation to gain insight on symbionts Haber M, Carbone M, Mollo E, Gavagnin M, Ilan M (2011) vs. transient microbes in high and low microbial abun- Chemical defense against predators and bacterial foul- dance sponges. Environ Microbiol 15:3008−3019 ing in the Mediterranean sponges Axinella polypoides Blunt JW, Copp BR, Hu WP, Munro MH, Northcote PT, Prin- and A. verrucosa. Mar Ecol Prog Ser 422: 113−122 sep MR (2009) Marine natural products. Nat Prod Rep 26: Hardoim CCP, Costa R (2014) Microbial communities and 170−244 bioactive compounds in marine sponges of the family Boudouresque CF, Verlaque M (2007) Ecology of Paracen- Irciniidae—a review. Mar Drugs 12: 5089−5122 trotus lividus. Dev Aquacult Fish Sci 37: 243−285 Hay ME, Kappel QE, Fenical W (1994) Synergisms in plant Boury-Esnault N, Rützler K (eds) (1997) Thesaurus of sponge defenses against herbivores: interactions of chemistry, morphology. Smithson Contrib Zool 596:1−55 calcification, and plant quality. Ecology 75:1714−1726 Burns E, Ilan M (2003) Comparison of anti-predatory de - Hill MS, Lopez NA, Young KA (2005) Anti-predator de- fenses of Red Sea and sponges. II. Physical fenses in western North Atlantic sponges with evidence defense. Mar Ecol Prog Ser 252:115−123 of enhanced defense through interactions between Buss LW (1976) Better living through chemistry: the relation- spicules and chemicals. Mar Ecol Prog Ser 291: 93−102 ship between allelochemical interactions and competitive Horn MH (1989) Biology of marine herbivorous fishes. networks. In: Harrison FW, Cowden RR (eds) Aspects Oceanogr Mar Biol Annu Rev 27: 167−272 of sponge biology. Academic Press, New York, NY, Johnston IS, Hildeman WH (1982) Cellular organization in p 315−327 the marine demosponge Callyspongia diffusa. Mar Biol Chanas B, Pawlik JR (1995) Defenses of Caribbean sponges 67: 1−7 against predatory reef fish. II. Spicules, tissue toughness, Jones AC, Blum JE, Pawlik JR (2005) Testing for defensive and nutritional quality. Mar Ecol Prog Ser 127:195−211 synergy in Caribbean sponges: bad taste or glass Chanas B, Pawlik JR (1996) Does the skeleton of a sponge spicules? J Exp Mar Biol Ecol 322:67−81 provide a defense against predatory reef fish? Oecologia Kabasakal H (2001) Description of the feeding morphology 107: 225−231 and the food habits of four sympatric labrids (Perci- Chesson PL (2000) Mechanisms of maintenance of species formes, Labridae) from south-eastern Aegean Sea, diversity. Annu Rev Ecol Syst 31: 343−366 Turkey. Neth J Zool 51: 439−455 Chesson PL, Warner RR (1981) Environmental variability Knowlton AL, Highsmith RC (2005) Nudibranch-sponge promotes coexistence in lottery competitive systems. Am feeding dynamics: benefits of symbiont-containing sponge Nat 117: 923−943 to Archidoris montereyensis (Cooper, 1862) and recovery 92 Mar Ecol Prog Ser 523: 81–92, 2015

of nudibranch feeding scars by Halichondria panicea Sokal RR, Rohlf FJ (1995) Biometry. Freeman & Co, New (Pallas, 1766). J Exp Mar Biol Ecol 327: 36−46 York, NY Lemee R, Boudouresque CF, Gobert J, Malestroit P, Mari X, Stergiou KI, Karpouzi VS (2002) Feeding habits and trophic Meinesz A, Ruitton S (1996) Feeding behaviour of Para- levels of Mediterranean fish. Rev Fish Biol Fish 11: centrotus lividus in the presence of Caulerpa taxifolia 217−254 introduced in the Mediterranean Sea. Oceanol Acta 19: Sureda A, Box A, Enseñat M, Alou E, Tauler P, Deudero S, 245−253 Pons A (2006) Enzymatic antioxidant response of a Leon YM, Bjorndal KA (2002) Selective feeding in the labrid fish (Coris julis) liver to environmental cauler- hawksbill turtle, an important predator in coral reef penyne. Comp Biochem Physiol C Toxicol Pharmacol ecosystems. Mar Ecol Prog Ser 245:249−258 144: 191−196 Leys SP, Reiswig HM (1998) Transport pathways in the- Taylor MW, Radax R, Steger D, Wagner M (2007) Sponge- neotropical sponge Aplysina. Biol Bull 195:30−42 associated microorganisms: evolution, ecology, and Loh TL, Pawlik JR (2014) Chemical defenses and resource biotechnological potential. Microbiol Mol Biol Rev 71: trade-offs structure sponge communities on Caribbean 295−347 coral reefs. Proc Natl Acad Sci USA 111: 4151−4156 Thacker RW, Freeman CJ (2012) Sponge−microbe sym- Marin A, Belluga ML, Scognamiglio G, Cimino G (1997) bioses: recent advances and new directions. Adv Mar Morphological and chemical camouflage of the Mediter- Biol 62: 57−111 ranean nudibranch Discodoris indecora on the sponges Thacker RW, Becerro MA, Lumbang WA, Paul VJ (1998) Ircinia variabilis and Ircinia fasciculata. J Molluscan Stud Allelopathic interactions between sponges on a tropical 63: 431−439 reef. Ecology 79: 1740−1750 Martin D, Uriz MJ (1993) Chemical bioactivity of Mediter- Uriz MJ (2006) Mineral skeletogenesis in sponges. Can J ranean benthic organisms against embryos and larvae of Zool 86: 1−35 marine invertebrates. J Exp Mar Biol Ecol 173: 11−27 Uriz MJ, Martin D, Rosell D (1992) Relationships of biologi- McClintock JB, Baker BJ, Slattery M, Hamann M, Kopitzke cal and taxonomic characteristics to chemically mediated R, Heine J (1994) Chemotactic tube-foot responses of bioactivity in Mediterranean littoral sponges. Mar Biol a spongivorous sea star Perknaster fuscus to organic 113: 287−297 extracts from antarctic sponges. J Chem Ecol 20: 859−870 Uriz MJ, Turon X, Becerro MA, Galera J (1996) Feeding Monks NR, Lerner C, Henriques AT, Farias FM, Schapoval deterrence in sponges. The role of toxicity, physical EE, Suyenaga ES, Mothes B (2002) Anticancer, anti - defenses, energetic contents, and life-history stage. chemotactic and antimicrobial activities of marine sponges J Exp Mar Biol Ecol 205: 187−204 collected off the coast of Santa Catarina, southern Brazil. Uriz MJ, Turon X, Becerro MA, Agell G (2003) Siliceous J Exp Mar Biol Ecol 281: 1−12 spicules and skeleton frameworks in sponges: origin, Pallaoro A, Santic M, Jardas I (2003) Feeding habits of the diversity, ultrastructural patterns, and biological func- saddled bream, Oblada melanura (Sparidae), in the Adri- tions. Microsc Res Tech 62:279−299 atic Sea. Cybium 27:261−268 Uriz MJ, Agell G, Blanquer A, Turon X, Casamayor EO Pallaoro A, Santic M, Jardas I (2006) Feeding habits of (2012) Endosymbiotic calcifying bacteria: a new cue to the common two-banded sea bream, Diplodus vulgaris the origin of calcification in Metazoa? Evolution 66: (Sparidae), in the eastern Adriatic Sea. Cybium 30: 19−25 2993−2999 Pennings SC, Svedberg JM (1993) Does CaCO3 in food deter van Alstyne KL, Paul VJ (1992) Chemical and structural feeding by sea urchins? Mar Ecol Prog Ser 101:163−167 defenses in the sea fan Gorgonia ventalina: effects Piel J, Hui D, Wen G, Butzke D, Platzer M, Fusetani N, against generalist and specialist predators. Coral Reefs Matsunaga S (2004) Antitumor polyketide biosynthesis 11: 155−159 by an uncultivated bacterial symbiont of the marine Van Soest R (2002) Family Hymedesmiidae Topsent, 1928. sponge Theonella swinhoei. Proc Natl Acad Sci USA 101: In: Hooper JNA, Van Soest RWM (eds) Systema Porifera: 16222−16227 a guide to the classification of sponges. Kluwer Aca- Ribeiro SM, Bianco ÉM, Rogers R, Teixeira VL, Pereira RC demic/Plenum Publishers, New York, NY, p 575−601 (2010) Chemical defense of Hymeniacidon heliophila Vergés A, Becerro MA, Alcoverro T, Romero J (2007) Exper- (Porifera: Halichondrida) against tropical predators. Braz imental evidence of chemical deterrence against multi- J Oceanogr 58:315−321 ple herbivores in the seagrass Posidonia oceanica. Mar Ribeiro SM, Cassiano KM, Cavalcanti DN, Teixeira VL, Ecol Prog Ser 343: 107−114 Pereira RC (2012) Isolated and synergistic effects of Verlaque M (1984) Biologie des juveniles de l’oursin her - chemical and structural defenses of two species of bivore Paracentrotus lividus Lamarck: séléctivité du Tethya (Porifera: Demospongiae). J Sea Res 68: 57−62 broutage et impact de l’éspece sur les communautés Rohde S, Schupp PJ (2011) Allocation of chemical and struc- algales de substract rocheux en Corse (Méditerranée, tural defenses in the sponge Melophlus sarasinorum. France). Bot Mar 27: 401−424 J Exp Mar Biol Ecol 399: 76−83 Verlaque M, Nédelec H (1983) Biologie de Paracentrotus Santos CP, Coutinho AB, Hajdu E (2002) Spongivory by lividus (Lamarck) sur substract rocheux en Corse (Médi- tribuloides from Salvador, Bahia (Echino - terranée, France): alimentation des adultes. Vie Milieu dermata: Echinoidea). J Mar Biol Assoc UK 82:295−297 33: 191−202 Schupp PJ, Paul VJ (1994) Calcification and secondary Wulff JL (2000) Sponge predators may determine differ- metabolites in tropical seaweeds: variable effects on ences in sponge fauna between two sets of mangrove herbivorous fishes. Ecology 75: 1172−1185 cays, Belize barrier reef. Atoll Res Bull 477: 251−263 Sipkema D, Franssen MCR, Osinga R, Tramper J, Wijffels Yekutieli D, Benjamini Y (1999) Resampling-based false RH (2005) Marine sponges as pharmacy. Mar Biotechnol discovery rate controlling multiple test procedures for 7: 142−162 correlated test statistics. J Stat Plan Inference 82: 171−196

Editorial responsibility: Antonio Bode, Submitted: June 11, 2014; Accepted: November 3, 2014 A Coruña, Spain Proofs received from author(s): February 28, 2015